feat: add engineering evaluation review and adoption UI

This commit is contained in:
Agent Workbench Dev
2026-11-01 09:56:25 +08:00
parent 2a44f15693
commit 17c9f9d0cd
16 changed files with 491 additions and 123 deletions
+6
View File
@@ -52,3 +52,9 @@ docs/desktop-host-*.json -text whitespace=blank-at-eol,blank-at-eof,space-before
docs/desktop-credentials-*.tap binary
docs/desktop-credentials-*.txt binary
docs/desktop-credentials-*.json -text whitespace=blank-at-eol,blank-at-eof,space-before-tab,cr-at-eol
docs/engineering-evals-*-evidence/** -text whitespace=blank-at-eol,blank-at-eof,space-before-tab,cr-at-eol
docs/engineering-evals-*-evidence/**/*.tap binary
docs/engineering-evals-*-evidence/*.tap binary
docs/engineering-evals-*-evidence/**/*.log binary
docs/engineering-evals-*-evidence/*.log binary
docs/engineering-evals-*.json -text whitespace=blank-at-eol,blank-at-eof,space-before-tab,cr-at-eol
+9 -1
View File
@@ -7,7 +7,7 @@
| A 固定契约 | 专项通过 | 1–3固定样本、HEAD/同源字节、明确源码/原测试/检查命令、模型和总预算 |
| B 运行账本 | 专项通过 | 原版0模型请求、候选原工具/审批/预算、停止与重启未知、不重放已尝试准备 |
| C 比较 | 专项通过 | 原失败/全部尝试/日志SHA、原用量缺失及人工审批/应用、不可变报告和父审查/Git门控 |
| D 产品流程 | 待实现 | 桌面/390px准备、审查、比较和采纳 |
| D 产品流程 | 7组通过 | 桌面/390px准备、审查、比较、导出、刷新、采纳、停止和来源变化;两视口实际截图回看 |
| E 实际工程 | 待验 | Node/React/Python及自身模块小范围试点 |
| F 完整发行 | 待验 | 完整源码/历史UI/独立源码/便携/安装/升级/原生回看 |
@@ -22,3 +22,11 @@
最后原版日志/采纳/HTTP3条路径3/3通过(`.qa/evaluation-abc-final-guards.log`)。首次选择器错误后还发现受控textarea的标签名称会随内容变化,现全部关键textarea均有稳定的aria-label,另开收据复验;原第二次失败`.qa/engineering-evals-ui-b08b6c81-cd2b-47c5-85d4-66ecba9fd922`保持。
父工程检查/变更集/Git/工作区各操作的活动门控已纳入评测活动,最终相关2路径2/2通过。A-C原日志按原字节封存于[清单](engineering-evals-abc-evidence/manifest.json)。
## D 产品流程回顾
最终7组通过,原收据`.qa/engineering-evals-ui-7203e02f-ee05-41ab-81dd-40b215086863`,概览见[实际UI记录](engineering-evals-ui-validation.json)。实际原版3测试失败、候选3测试通过、父审查/检查/Git均完成;停止保留stopped,父来源变化在创建工作区前拒绝。桌面并列和390px堆叠截图已目视回看,原失败与候选通过、人工审批/应用次数、缺失费用及固定响应标签均可读,未见页面横向溢出。
第三次UI收据`.qa/engineering-evals-ui-d240669e-0c9b-4450-960e-1b870f9fb23f`前5组已通过,之后QA在POST完成前读取最新批次,误选旧批次导致停止场景等待失败。修正为等待批次数量和新标题后获取原对象;原失败保持,最终轮独立执行完整7组。此为验收脚本竞态,不删除或改写原失败。
A-C原日志在提交时被Git默认换行规范化;现增加证据目录字节保护,并重新按原`.qa`字节暂存,后续提交逐Git blob核对清单SHA。保留旧提交,不改写历史。
+6
View File
@@ -17,3 +17,9 @@
复用了原WorkerManager/WorkspaceManager、CheckRunner、ChangeSet及GitDelivery。原版不伪装成一个有修改的worker成果,评测归集只接受候选;普通协作接口不能绕过评测开始或采纳。全部检查尝试、原日志SHA及上游未知用量进入固定报告,缺失证据显示unknown并阻止采纳,不删除原分母。多样本结束后只能采纳一项;先核对原父来源,再切回所属Task形成独立变更集,不能直接覆盖父文件。
回顾发现采纳预检含异步窗口,已增加父工程活动锁和持久采纳意图;重启不恢复副作用。计划停止也等待原准备结束。对源码/模型修订、精确文件、原测试、HTTP归属及停止恢复的12项专项和原worker6项合计18/18通过,原10项首轮及后续修正证据分别保留。工程推理成功率、真实模型费用和Purge效果没有据此判为完成。
## D 回顾
实际原任务导航、提案审查、检查审批、对照报告与采纳后父交付7组通过。对照面板将固定响应标签放在报告头部,390px按原版/候选顺序堆叠并实际滚动核对候选测试证据;缺失费用显示未报告。修正可访问名称随select选项/textarea内容变化的问题,关键表单有稳定名称。第三轮QA提前读取最新对象的竞态修正后,第四轮完整通过;原失败收据保持。复用了现有模态、WorkerPanel、差异审查和Git交付,没有新增执行内核或依赖。
回看补充同批切换样本时迟到报告响应的当前样本身份核对,避免旧报告盖住新选择;TypeScript通过,随后完整历史/新评测回归覆盖最终代码。证据目录保留原CRLF字节并显式声明cr-at-eol,避免Git文本规范化改变日志及清单。
@@ -1,18 +1,18 @@
✔ evaluation contract fixes bounded samples, source paths, checks and aggregate budget (525.0705ms)
✔ same-source workspaces seal real commands before execution and preserve original staged and unstaged bytes (11083.0145ms)
✔ actual original failures and candidate success are both preserved, then adoption requires parent review, checks and Git (40331.2376ms)
✔ candidate cannot shrink tests and changing original test bytes rejects execution before any model request (15626.3462ms)
✔ missing upstream usage remains unknown and does not become a free run or invented cost (25914.3911ms)
✔ missing raw check logs preserve unknown result and original sample denominator (25759.3229ms)
✔ source and model changes never silently replace an existing batch input (2304.4183ms)
✔ sample queue cannot skip or reset attempted samples, and reports freeze further writes (20862.9625ms)
✔ stop cancels only current candidate and preserves partial results instead of replaying a request (16535.6083ms)
✔ restart recovers interrupted evaluations as unknown and keeps completed checks and plan identity (15402.608ms)
ℹ tests 10
ℹ suites 0
ℹ pass 10
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 174485.0587
✔ evaluation contract fixes bounded samples, source paths, checks and aggregate budget (525.0705ms)
✔ same-source workspaces seal real commands before execution and preserve original staged and unstaged bytes (11083.0145ms)
✔ actual original failures and candidate success are both preserved, then adoption requires parent review, checks and Git (40331.2376ms)
✔ candidate cannot shrink tests and changing original test bytes rejects execution before any model request (15626.3462ms)
✔ missing upstream usage remains unknown and does not become a free run or invented cost (25914.3911ms)
✔ missing raw check logs preserve unknown result and original sample denominator (25759.3229ms)
✔ source and model changes never silently replace an existing batch input (2304.4183ms)
✔ sample queue cannot skip or reset attempted samples, and reports freeze further writes (20862.9625ms)
✔ stop cancels only current candidate and preserves partial results instead of replaying a request (16535.6083ms)
✔ restart recovers interrupted evaluations as unknown and keeps completed checks and plan identity (15402.608ms)
ℹ tests 10
ℹ suites 0
ℹ pass 10
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 174485.0587
@@ -1,11 +1,11 @@
✔ actual original failures and candidate success are both preserved, then adoption requires parent review, checks and Git (42230.2742ms)
✔ missing raw check logs preserve unknown result and original sample denominator (31180.7936ms)
✔ HTTP evaluations require host token, original fingerprint and session ownership; worker API cannot bypass evaluation flow (15068.7458ms)
ℹ tests 3
ℹ suites 0
ℹ pass 3
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 88729.2571
✔ actual original failures and candidate success are both preserved, then adoption requires parent review, checks and Git (42230.2742ms)
✔ missing raw check logs preserve unknown result and original sample denominator (31180.7936ms)
✔ HTTP evaluations require host token, original fingerprint and session ownership; worker API cannot bypass evaluation flow (15068.7458ms)
ℹ tests 3
ℹ suites 0
ℹ pass 3
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 88729.2571
@@ -1,10 +1,10 @@
✔ actual original failures and candidate success are both preserved, then adoption requires parent review, checks and Git (41949.8076ms)
✔ HTTP evaluations require host token, original fingerprint and session ownership; worker API cannot bypass evaluation flow (12432.7767ms)
ℹ tests 2
ℹ suites 0
ℹ pass 2
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 54628.6602
✔ actual original failures and candidate success are both preserved, then adoption requires parent review, checks and Git (41949.8076ms)
✔ HTTP evaluations require host token, original fingerprint and session ownership; worker API cannot bypass evaluation flow (12432.7767ms)
ℹ tests 2
ℹ suites 0
ℹ pass 2
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 54628.6602
@@ -1,26 +1,26 @@
✔ evaluation contract fixes bounded samples, source paths, checks and aggregate budget (879.7691ms)
✔ same-source workspaces seal real commands before execution and preserve original staged and unstaged bytes (14874.6757ms)
✔ actual original failures and candidate success are both preserved, then adoption requires parent review, checks and Git (49362.4978ms)
✔ candidate cannot shrink tests and changing original test bytes rejects execution before any model request (19490.2731ms)
✔ missing upstream usage remains unknown and does not become a free run or invented cost (28035.4922ms)
✔ missing raw check logs preserve unknown result and original sample denominator (27592.2318ms)
✔ source and model changes never silently replace an existing batch input (2375.6565ms)
✔ model revision change before preparing a fixed sample cannot silently use the replacement connection (2210.0987ms)
✔ HTTP evaluations require host token, original fingerprint and session ownership; worker API cannot bypass evaluation flow (12196.9645ms)
✔ sample queue cannot skip or reset attempted samples, and reports freeze further writes (21512.0204ms)
✔ stop cancels only current candidate and preserves partial results instead of replaying a request (17772.6846ms)
✔ restart recovers interrupted evaluations as unknown and keeps completed checks and plan identity (16171.5361ms)
✔ worker scopes and allocated budgets reject aliases, extra permissions and aggregate overflow (770.0996ms)
✔ two sequential Git workspaces freeze a new parent goal, separate owned tasks and preserve all source bytes (13643.8712ms)
✔ real isolated changes, approvals and npm checks produce original owned outputs, not model completion claims (28402.4151ms)
✔ forged worker tools cannot execute terminal or Git even when absent from the advertised tool list (12969.4039ms)
✔ a completion-only model pauses each worker without fabricated changes or checks (14462.927ms)
✔ restart retains frozen scope and marks interrupted parent and workers unknown without sending requests (12239.9837ms)
ℹ tests 18
ℹ suites 0
ℹ pass 18
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 212727.8273
✔ evaluation contract fixes bounded samples, source paths, checks and aggregate budget (879.7691ms)
✔ same-source workspaces seal real commands before execution and preserve original staged and unstaged bytes (14874.6757ms)
✔ actual original failures and candidate success are both preserved, then adoption requires parent review, checks and Git (49362.4978ms)
✔ candidate cannot shrink tests and changing original test bytes rejects execution before any model request (19490.2731ms)
✔ missing upstream usage remains unknown and does not become a free run or invented cost (28035.4922ms)
✔ missing raw check logs preserve unknown result and original sample denominator (27592.2318ms)
✔ source and model changes never silently replace an existing batch input (2375.6565ms)
✔ model revision change before preparing a fixed sample cannot silently use the replacement connection (2210.0987ms)
✔ HTTP evaluations require host token, original fingerprint and session ownership; worker API cannot bypass evaluation flow (12196.9645ms)
✔ sample queue cannot skip or reset attempted samples, and reports freeze further writes (21512.0204ms)
✔ stop cancels only current candidate and preserves partial results instead of replaying a request (17772.6846ms)
✔ restart recovers interrupted evaluations as unknown and keeps completed checks and plan identity (16171.5361ms)
✔ worker scopes and allocated budgets reject aliases, extra permissions and aggregate overflow (770.0996ms)
✔ two sequential Git workspaces freeze a new parent goal, separate owned tasks and preserve all source bytes (13643.8712ms)
✔ real isolated changes, approvals and npm checks produce original owned outputs, not model completion claims (28402.4151ms)
✔ forged worker tools cannot execute terminal or Git even when absent from the advertised tool list (12969.4039ms)
✔ a completion-only model pauses each worker without fabricated changes or checks (14462.927ms)
✔ restart retains frozen scope and marks interrupted parent and workers unknown without sending requests (12239.9837ms)
ℹ tests 18
ℹ suites 0
ℹ pass 18
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 212727.8273
@@ -1,48 +1,48 @@
{
"stage": "1.20 A-C",
"first": {
"passed": 10,
"failed": 0
},
"reviewed": {
"evaluations": 12,
"workers": 6,
"passed": 18,
"failed": 0
},
"finalGuardPaths": {
"passed": 3,
"failed": 0
},
"finalOperationLockPaths": {
"passed": 2,
"failed": 0
},
"realModelRequests": 0,
"files": [
{
"source": ".qa\\evaluation-a-first.log",
"file": "evaluation-a-first.tap",
"bytes": 1259,
"sha256": "8d5ba79071a3339a5ac2affdcb13a6af88f64fc9c4932c209bca53a0da6b3caf"
},
{
"source": ".qa\\evaluation-abc-reviewed.log",
"file": "evaluation-abc-reviewed.tap",
"bytes": 2228,
"sha256": "15b798551aa6dee8e5b12d73fc070272d1c67b6ffdbc6445d41fcb63d95ec8fa"
},
{
"source": ".qa\\evaluation-abc-final-guards.log",
"file": "evaluation-abc-final-guards.tap",
"bytes": 504,
"sha256": "2db5ee72d4439acb97176bedb14985ffe91cc7ecbc361e32a969b08b39a340f9"
},
{
"source": ".qa\\evaluation-abc-operation-locks.log",
"file": "evaluation-abc-operation-locks.tap",
"bytes": 405,
"sha256": "f689978a90d947c0561dc8f5a891b51b432b547dc206fe302279566957a6c75d"
}
]
}
{
"stage": "1.20 A-C",
"first": {
"passed": 10,
"failed": 0
},
"reviewed": {
"evaluations": 12,
"workers": 6,
"passed": 18,
"failed": 0
},
"finalGuardPaths": {
"passed": 3,
"failed": 0
},
"finalOperationLockPaths": {
"passed": 2,
"failed": 0
},
"realModelRequests": 0,
"files": [
{
"source": ".qa\\evaluation-a-first.log",
"file": "evaluation-a-first.tap",
"bytes": 1259,
"sha256": "8d5ba79071a3339a5ac2affdcb13a6af88f64fc9c4932c209bca53a0da6b3caf"
},
{
"source": ".qa\\evaluation-abc-reviewed.log",
"file": "evaluation-abc-reviewed.tap",
"bytes": 2228,
"sha256": "15b798551aa6dee8e5b12d73fc070272d1c67b6ffdbc6445d41fcb63d95ec8fa"
},
{
"source": ".qa\\evaluation-abc-final-guards.log",
"file": "evaluation-abc-final-guards.tap",
"bytes": 504,
"sha256": "2db5ee72d4439acb97176bedb14985ffe91cc7ecbc361e32a969b08b39a340f9"
},
{
"source": ".qa\\evaluation-abc-operation-locks.log",
"file": "evaluation-abc-operation-locks.tap",
"bytes": 405,
"sha256": "f689978a90d947c0561dc8f5a891b51b432b547dc206fe302279566957a6c75d"
}
]
}
+73
View File
@@ -0,0 +1,73 @@
{
"passed": true,
"version": "1.19.0",
"evidenceKind": "owned fixed response, actual isolated Git/checks and desktop/390px UI",
"realModelRequests": 0,
"fixtureRequests": 6,
"assertions": [
"desktop/390px form rejects candidate test edits, keeps entered requirements, persists fixed model/source/budget and reloads the original plan",
"explicit preparation seals two same-source checkouts and actual commands; original real three-test failure is retained with zero model requests",
"candidate navigates to its original restricted task for real change review and check approval, runs the original test/build and preserves the parent source",
"parallel comparison preserves original failure, candidate pass, real approval/application counts and absent cost; export and refresh retain the original report and fixture label",
"adoption forms a fresh parent review batch; completion requires applied source, fresh real parent checks and local Git delivery while owner modifications remain exact",
"stop waits for the owned candidate request, preserves stopped outcome and never turns cancellation into a passed sample or a replay",
"changed owner source is rejected before workspace creation and the error stays attached to the original batch instead of a generic chat reply"
],
"screenshots": [
{
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\engineering-evals-ui-7203e02f-ee05-41ab-81dd-40b215086863\\desktop-empty.png",
"sha256": "a2180342db5c91cca02d41558019c334b83314eba89863005123813d68cce19e"
},
{
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\engineering-evals-ui-7203e02f-ee05-41ab-81dd-40b215086863\\mobile-form-scope-error.png",
"sha256": "79ff7a3dd7a3e016cbc61ded4e41f7298085fbd0a7256ba0d5d6aa15c403f7ac"
},
{
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\engineering-evals-ui-7203e02f-ee05-41ab-81dd-40b215086863\\desktop-ready.png",
"sha256": "743a33128dbf82a517bcda9f116b73c6b3ac8f0a800747c06561c60efc275774"
},
{
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\engineering-evals-ui-7203e02f-ee05-41ab-81dd-40b215086863\\desktop-candidate-complete.png",
"sha256": "536cb42c8840112a2069682624428bcecb6ab79863b2296bf5542f89c86faffd"
},
{
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\engineering-evals-ui-7203e02f-ee05-41ab-81dd-40b215086863\\desktop-comparison.png",
"sha256": "12089849e6fc1113946abcd24c5208a6413a7d8f60973d4bd6d7466cbe0eac0c"
},
{
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\engineering-evals-ui-7203e02f-ee05-41ab-81dd-40b215086863\\mobile-comparison.png",
"sha256": "0116a8965610d0e2b6a12e6b7ef9f0b46af530dd31e01ea14f2cd50394f585d0"
},
{
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\engineering-evals-ui-7203e02f-ee05-41ab-81dd-40b215086863\\mobile-candidate-evidence.png",
"sha256": "594bbd2c3d8171c3f8e0aa8fadd9f2339fdb4bac8cd6827545433ed81aa21433"
},
{
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\engineering-evals-ui-7203e02f-ee05-41ab-81dd-40b215086863\\desktop-delivered.png",
"sha256": "d67cc3eb5dbaccf3b94dede4f487dd75acf3b071651958362444e9f63107df49"
},
{
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\engineering-evals-ui-7203e02f-ee05-41ab-81dd-40b215086863\\mobile-stopped.png",
"sha256": "6f69b1a0313d9dc7ab33ce21b4a35e83f89a4157be825fe2f753738e20fb6e22"
},
{
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\engineering-evals-ui-7203e02f-ee05-41ab-81dd-40b215086863\\mobile-source-error.png",
"sha256": "926f1c3cafb7690828d7cf1001dfa6d28046895b053897675467b937da9b6aad"
}
],
"exports": [
{
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\engineering-evals-ui-7203e02f-ee05-41ab-81dd-40b215086863\\downloaded-report.json",
"sha256": "fdb56d22b1b5f9778fba08c2ceb0ce57f5d61aed2d19b8ef7f37028c90c895a2"
},
{
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\engineering-evals-ui-7203e02f-ee05-41ab-81dd-40b215086863\\report-8a1d20f4-04ef-421e-8939-342e5422abd0.json",
"sha256": "fdb56d22b1b5f9778fba08c2ceb0ce57f5d61aed2d19b8ef7f37028c90c895a2"
},
{
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\engineering-evals-ui-7203e02f-ee05-41ab-81dd-40b215086863\\report-606d02f2-67b9-443c-bf44-4b36fd764a26.json",
"sha256": "ba603f2c9c8fbdb7ad116e4fcedc17762576258f084179c0bee7bb8cf6586c0c"
}
],
"errors": []
}
+79
View File
@@ -0,0 +1,79 @@
import fs from 'node:fs';
import path from 'node:path';
import assert from 'node:assert/strict';
import { randomUUID, createHash } from 'node:crypto';
import { chromium, expect } from 'playwright/test';
import { fixture, call } from '../tests/helpers.mjs';
import { workerHttp } from '../tests/fixtures/workers-http.mjs';
import { ownerBytes } from '../tests/fixtures/workers-project.mjs';
if(process.argv.includes('--help')){console.log('node scripts/verify-evaluations-ui.mjs — independent local HTTP host, fixed-response candidate, real Git/test/build, desktop/390px queue, review, comparison, adoption, stop and source errors. No real model inference.');process.exit(0);}
assert.equal(process.argv.length,2);
const directory=path.resolve('.qa','engineering-evals-ui-'+randomUUID());fs.mkdirSync(directory,{recursive:true});
const disposers=[],t={after:fn=>disposers.push(fn)},f=fixture(t),hash=bytes=>createHash('sha256').update(bytes).digest('hex');
let hold=false;
const host=await workerHttp(f,{broken:true,respond:async({project,signal})=>{
if(hold)return new Promise((resolve,reject)=>{if(signal.aborted)reject(signal.reason);else signal.addEventListener('abort',()=>reject(signal.reason),{once:true});});
const owner=[...f.store.sessions.values()].find(session=>session.projectId===project.id),taskId=owner.taskId;
const plan=[...f.store.core.check_plan.values()].find(plan=>plan.taskId===taskId),job=[...f.store.core.check_job.values()].find(job=>job.taskId===taskId);
let calls=[];
if(!owner.changes.length)calls=[call(randomUUID(),'propose_file',{path:'src/add.mjs',content:'export const add = (a,b) => a + b;\n',reason:'本地固定响应修复加法'})];
else if(owner.changes.some(change=>change.status==='pending'))calls=[];
else if(!plan)calls=[call(randomUUID(),'plan_checks',{names:['test','build']})];
else if(!job)calls=[call(randomUUID(),'run_checks',{plan_id:plan.id})];
return {message:{role:'assistant',content:calls.length?'':'固定响应已返回,仍需实际工程证据',...(calls.length?{tool_calls:calls}:{})},usage:{total_tokens:168,prompt_tokens:123,completion_tokens:45}};
}});
const browser=await chromium.launch({channel:'msedge',headless:true}),page=await browser.newPage({viewport:{width:1440,height:1000},acceptDownloads:true});
const assertions=[],screenshots=[],errors=[],exports=[];page.on('pageerror',error=>errors.push(error.message));
const panel=()=>page.getByLabel('工程评测控制台',{exact:true});
const mark=value=>{assertions.push(value);console.log('PASS '+value);};
const shot=async name=>{const file=path.join(directory,name+'.png');await page.screenshot({path:file,fullPage:true});screenshots.push({file,sha256:hash(fs.readFileSync(file))});assert.ok(await page.locator('body').evaluate(node=>node.scrollWidth<=innerWidth),'body overflow '+name);assert.ok(await panel().evaluate(node=>node.scrollWidth<=node.clientWidth+1),'panel overflow '+name);};
const open=async()=>{await page.getByRole('button',{name:'工程评测',exact:true}).click();await expect(panel().getByRole('button',{name:'刷新',exact:true})).toBeEnabled();};
const confirm=async(name,button)=>{await panel().getByRole('button',{name,exact:true}).click();await page.getByRole('dialog',{name:'确认评测操作',exact:true}).getByRole('button',{name:button,exact:true}).click();};
const refresh=async()=>{await panel().getByRole('button',{name:'刷新',exact:true}).click();};
const latest=()=>[...f.store.core.evaluation.values()].at(-1);
const parent=async()=>{await page.getByRole('button',{name:'隔离协作',exact:true}).click();await page.getByLabel('隔离协作控制台',{exact:true}).getByRole('button',{name:'返回父目标',exact:true}).click();await expect.poll(()=>page.evaluate(()=>localStorage.getItem('awb-session'))).toBe(f.session.id);await open();};
async function fill(title){await panel().getByLabel('评测名称',{exact:true}).fill(title);await panel().getByLabel('评测来源',{exact:true}).selectOption('local-fixture');await panel().getByLabel('样本名称',{exact:true}).fill('正负数和整数加法边界');await panel().getByLabel('具体需求',{exact:true}).fill('修复加法运算,保留原测试及已有调用方式');await panel().getByLabel('验收条件',{exact:true}).fill('原3项加法测试与构建通过,原测试文件和运行声明保持不变');await panel().getByLabel('允许修改的源码文件',{exact:true}).fill('src/add.mjs');}
async function apply(prepared=false){if(await page.getByRole('button',{name:'查看改动',exact:true}).isVisible())await page.getByRole('button',{name:'查看改动',exact:true}).click();else await page.locator('.inspector-tabs').getByRole('button',{name:/^改动/}).click();await page.getByRole('button',{name:prepared?'审查变更集':'组合审查',exact:true}).click();if(!prepared)await page.getByRole('dialog',{name:'选择要一起审查的提案',exact:true}).getByRole('button',{name:'封存审查范围',exact:true}).click();const dialog=page.getByRole('dialog',{name:'多文件变更集审查',exact:true});await dialog.getByRole('button',{name:'确认应用全部 1 项',exact:true}).click();await expect(dialog).toContainText('全部应用完成',{timeout:20000});await page.keyboard.press('Escape');}
let result;
try{
await page.addInitScript(({projectId,sessionId})=>{localStorage.setItem('awb-project',projectId);localStorage.setItem('awb-session',sessionId);},{projectId:f.project.id,sessionId:f.session.id});
await page.goto(host.url);await page.waitForLoadState('networkidle');await open();await shot('desktop-empty');
await expect(panel().getByRole('button',{name:'封存评测计划',exact:true})).toBeDisabled();await fill('固定工程改进 · 中文完整评测');
await panel().getByLabel('允许修改的源码文件',{exact:true}).fill('test/add.test.mjs');await panel().getByRole('button',{name:'封存评测计划',exact:true}).click();await expect(panel().getByRole('alert')).toContainText('测试');await expect(panel().getByLabel('具体需求',{exact:true})).toHaveValue('修复加法运算,保留原测试及已有调用方式');
await page.setViewportSize({width:390,height:844});await shot('mobile-form-scope-error');await page.setViewportSize({width:1440,height:1000});
await panel().getByLabel('允许修改的源码文件',{exact:true}).fill('src/add.mjs');await panel().getByRole('button',{name:'封存评测计划',exact:true}).click();await expect(panel()).toContainText('计划已封存');
const batch=latest(),sample=batch.samples[0],original=ownerBytes(f);assert.equal(batch.evidenceMode,'local-fixture');assert.equal(host.requests.length,0);
await page.reload();await page.waitForLoadState('networkidle');await open();await expect(panel()).toContainText(batch.title);
mark('desktop/390px form rejects candidate test edits, keeps entered requirements, persists fixed model/source/budget and reloads the original plan');
await panel().getByRole('button',{name:'准备同源工作区',exact:true}).click();await page.keyboard.press('Escape');assert.equal(f.store.core.worker.size,0);
await confirm('准备同源工作区','确认准备');await expect.poll(()=>sample.status,{timeout:30000}).toBe('ready');await expect(panel()).toContainText('node --test test/*.test.mjs');assert.deepEqual(ownerBytes(f),original);await shot('desktop-ready');
await confirm('核对并运行原版','确认运行原版');await expect.poll(()=>sample.status,{timeout:30000}).toBe('baseline_complete');assert.equal(host.requests.length,0);assert.equal(f.store.core.check_job.get(sample.baselineJobId).results[0].tests.failed,3);
mark('explicit preparation seals two same-source checkouts and actual commands; original real three-test failure is retained with zero model requests');
await confirm('开始候选改进','确认开始候选');await expect.poll(()=>sample.status,{timeout:30000}).toBe('awaiting_review');
await panel().getByRole('button',{name:'审查候选改动',exact:true}).click();await apply();await parent();
await confirm('继续候选任务','确认开始候选');const group=f.store.core.collaboration.get(sample.groupId),child=group.children.find(child=>child.evaluation.role==='candidate');
await expect.poll(()=>f.store.session(child.sessionId).status,{timeout:15000}).toBe('awaiting_approval');await expect(panel().getByRole('button',{name:'核对并审批检查',exact:true})).toBeVisible();
await panel().getByRole('button',{name:'核对并审批检查',exact:true}).click();await page.locator('.approval-card').getByRole('button',{name:'确认执行',exact:true}).click();await parent();
await expect.poll(()=>sample.status,{timeout:30000}).toBe('candidate_complete');assert.deepEqual(ownerBytes(f),original);await shot('desktop-candidate-complete');
mark('candidate navigates to its original restricted task for real change review and check approval, runs the original test/build and preserves the parent source');
await confirm('封存当前结果','确认封存结果');await expect.poll(()=>batch.status,{timeout:15000}).toBe('completed');await panel().getByRole('button',{name:'查看原版 / 候选对照',exact:true}).click();
const report=page.getByRole('dialog',{name:'原版与候选对照',exact:true});await expect(report).toContainText('测试 0/3 · 失败 3');await expect(report).toContainText('测试 3/3 · 失败 0');await expect(report).toContainText('不证明模型自主完成率');await shot('desktop-comparison');
const download=page.waitForEvent('download');await report.getByRole('button',{name:'导出原报告',exact:true}).click();const file=path.join(directory,'downloaded-report.json');await (await download).saveAs(file);const downloaded=JSON.parse(fs.readFileSync(file,'utf8'));assert.deepEqual(downloaded.summary,{original:'failed',candidate:'passed'});exports.push({file,sha256:hash(fs.readFileSync(file))});
await page.setViewportSize({width:390,height:844});await shot('mobile-comparison');await report.getByText('测试 3/3 · 失败 0',{exact:true}).scrollIntoViewIfNeeded();await expect(report.getByText('测试 3/3 · 失败 0',{exact:true})).toBeVisible();await shot('mobile-candidate-evidence');await page.keyboard.press('Escape');await page.reload();await page.waitForLoadState('networkidle');await open();await panel().getByRole('button',{name:'查看原版 / 候选对照',exact:true}).click();await expect(page.getByRole('dialog',{name:'原版与候选对照',exact:true})).toContainText('测试 0/3 · 失败 3');await page.keyboard.press('Escape');await page.setViewportSize({width:1440,height:1000});
mark('parallel comparison preserves original failure, candidate pass, real approval/application counts and absent cost; export and refresh retain the original report and fixture label');
await confirm('采纳此候选,准备父工程审查','确认准备采纳');await expect.poll(()=>batch.status,{timeout:30000}).toBe('awaiting_delivery');await panel().getByRole('button',{name:'审查父工程变更集',exact:true}).click();await apply(true);
const plan=await host.runtime.checkRunner().plan(f.session,['test','build']);assert.equal((await (await host.runtime.checkRunner().start(f.session,plan.id)).promise).status,'passed');
const delivery=await host.runtime.gitDelivery().plan(f.session,{message:'Adopt fixed engineering evaluation candidate'});assert.equal((await (await host.runtime.gitDelivery().start(f.session,delivery.id)).promise).status,'completed');await open();await panel().getByRole('button',{name:'核对父工程检查与 Git',exact:true}).click();await expect.poll(()=>batch.adoption.status).toBe('completed');assert.deepEqual(ownerBytes(f).files,original.files);await shot('desktop-delivered');
mark('adoption forms a fresh parent review batch; completion requires applied source, fresh real parent checks and local Git delivery while owner modifications remain exact');
hold=true;await panel().getByRole('button',{name:'新建下一批评测',exact:true}).click();await fill('停止中的候选,保留原失败分母');await panel().getByRole('button',{name:'封存评测计划',exact:true}).click();await expect.poll(()=>f.store.core.evaluation.size).toBe(2);await expect(panel()).toContainText('停止中的候选,保留原失败分母');const stopped=latest(),stoppedSample=stopped.samples[0];
await confirm('准备同源工作区','确认准备');await expect.poll(()=>stoppedSample.status,{timeout:30000}).toBe('ready');await confirm('核对并运行原版','确认运行原版');await expect.poll(()=>stoppedSample.status,{timeout:30000}).toBe('baseline_complete');const requests=host.requests.length;
await confirm('开始候选改进','确认开始候选');await expect.poll(()=>host.requests.length).toBe(requests+1);await panel().getByRole('button',{name:'停止本批执行',exact:true}).click();await expect.poll(()=>host.runtime.hasActivity(),{timeout:10000}).toBe(false);await refresh();assert.notEqual(stoppedSample.status,'candidate_complete');
await confirm('封存当前结果','确认封存结果');await expect.poll(()=>!!stoppedSample.report).toBe(true);assert.equal(stoppedSample.report.summary.candidate,'stopped');assert.equal(host.requests.length,requests+1);await page.setViewportSize({width:390,height:844});await shot('mobile-stopped');
mark('stop waits for the owned candidate request, preserves stopped outcome and never turns cancellation into a passed sample or a replay');
await panel().getByRole('button',{name:'新建下一批评测',exact:true}).click();await fill('来源变化时保持失效原因');await panel().getByRole('button',{name:'封存评测计划',exact:true}).click();await expect.poll(()=>f.store.core.evaluation.size).toBe(3);await expect(panel()).toContainText('来源变化时保持失效原因');const stale=latest();fs.appendFileSync(path.join(f.projectRoot,'README.md'),'\nOwner changed after plan\n');await confirm('准备同源工作区','确认准备');await expect.poll(()=>stale.samples[0].status,{timeout:10000}).toBe('failed');await expect(panel()).toContainText('来源工程已变化');await shot('mobile-source-error');
mark('changed owner source is rejected before workspace creation and the error stays attached to the original batch instead of a generic chat reply');
for(const batch of f.store.core.evaluation.values())for(const sample of batch.samples){if(sample.report){const file=path.join(directory,'report-'+sample.id+'.json');fs.writeFileSync(file,JSON.stringify(host.runtime.evaluationManager().readReport(f.session,batch,sample),null,2));exports.push({file,sha256:hash(fs.readFileSync(file))});}}
assert.deepEqual(errors,[]);result={passed:true,version:JSON.parse(fs.readFileSync('package.json','utf8')).version,evidenceKind:'owned fixed response, actual isolated Git/checks and desktop/390px UI',realModelRequests:0,fixtureRequests:host.requests.length,assertions,screenshots,exports,errors};
}catch(error){result={passed:false,error:error.stack,assertions,screenshots,exports,errors};await shot('failure').catch(()=>{});throw error;}
finally{await browser.close();await host.runtime.stopAll();f.store.storage.database.exec('PRAGMA wal_checkpoint(TRUNCATE)');fs.cpSync(f.root,path.join(directory,'original-fixture'),{recursive:true});fs.writeFileSync(path.join(directory,'result.json'),JSON.stringify(result,null,2));if(result.passed)fs.writeFileSync('docs/engineering-evals-ui-validation.json',JSON.stringify(result,null,2));for(const dispose of disposers.reverse())await dispose();console.log('EVIDENCE '+directory);}
+4
View File
@@ -14,6 +14,7 @@ import { ProjectMemoryPanel } from './ProjectMemoryPanel';
import { SkillCenter } from './SkillCenter';
import { McpPanel } from './McpPanel';
import { WorkerPanel } from './WorkerPanel';
import { EvaluationPanel } from './EvaluationPanel';
import { DesktopPanel } from './DesktopPanel';
import { CompanionPanel } from './CompanionPanel';
import { desktop } from './desktop';
@@ -82,6 +83,7 @@ export default function App() {
const [skillsOpen, setSkillsOpen] = useState(false);
const [mcpOpen, setMcpOpen] = useState(false);
const [workersOpen, setWorkersOpen] = useState(false);
const [evaluationsOpen, setEvaluationsOpen] = useState(false);
const [desktopOpen, setDesktopOpen] = useState(false);
const [companionOpen, setCompanionOpen] = useState(false);
const [pickingProject, setPickingProject] = useState(false);
@@ -548,6 +550,7 @@ export default function App() {
<IconButton label="项目MCP工具" onClick={() => setMcpOpen(true)} disabled={!selectedProject || restrictedWorker}><Puzzle size={15} /></IconButton>
<IconButton label="查看任务记录" onClick={() => showInspector('task')}><ClipboardCheck size={15} /></IconButton>
<IconButton label="隔离协作" onClick={() => setWorkersOpen(true)} disabled={!session}><GitBranch size={15} /></IconButton>
<IconButton label="工程评测" onClick={() => setEvaluationsOpen(true)} disabled={!session || restrictedWorker}><ClipboardCheck size={15} /></IconButton>
<IconButton label="本地代理引擎" onClick={() => setEngineOpen(true)} disabled={!session || restrictedWorker && !session.execution?.workers?.some(value => value.engine)}><Link2 size={15} /></IconButton>
<IconButton label="REA 工件研究" onClick={() => void openRea()} disabled={!selectedProject || creatingSession || restrictedWorker}><BookOpenText size={15} /></IconButton>
{narrow && <><IconButton label="查看文件" onClick={() => showInspector('file')}><FileCode2 size={16} /></IconButton><IconButton label="查看改动" onClick={() => showInspector('changes')}><GitBranch size={16} /></IconButton><IconButton label="查看执行轨迹" onClick={() => showInspector('trace')}><SquareTerminal size={16} /></IconButton></>}
@@ -598,6 +601,7 @@ export default function App() {
{skillsOpen && selectedProject && <Modal title="项目技能" description="管理当前项目的技能来源、版本与启用状态。" wide onClose={() => setSkillsOpen(false)}><SkillCenter key={selectedProject.id} projectId={selectedProject.id} busy={isRunning} /></Modal>}
{mcpOpen && selectedProject && <Modal title="项目 MCP 工具" description="实际探测本地 stdio 服务,选择可以请求的工具,并核对每次运行原记录。" wide onClose={() => setMcpOpen(false)}><McpPanel key={selectedProject.id + ':' + (session?.id ?? '')} projectId={selectedProject.id} session={session?.projectId === selectedProject.id ? session : null} busy={isRunning} onSession={acceptSession} onError={session ? failureFor(session) : failure} /></Modal>}
{workersOpen && session && <Modal title="隔离协作" description="固定父子目标、独立工作区、预算与原成果审查。" wide onClose={() => setWorkersOpen(false)}><WorkerPanel key={session.id} session={session} config={config} busy={isRunning} onSession={acceptSession} onNavigate={async (projectId, sessionId, view) => { setWorkersOpen(false); await navigateWorkspace(projectId, sessionId, () => true, view); }} onReview={() => { setWorkersOpen(false); showInspector('changes'); }} /></Modal>}
{evaluationsOpen && session && <Modal title="工程质量评测" description="固定需求与测试,比较原版和候选,审查后交付。" wide onClose={() => setEvaluationsOpen(false)}><EvaluationPanel key={session.id} session={session} config={config} busy={isRunning} onSession={acceptSession} onNavigate={async (projectId, sessionId, view) => { setEvaluationsOpen(false); await navigateWorkspace(projectId, sessionId, () => true, view); }} onReview={() => { setEvaluationsOpen(false); showInspector('changes'); }} onRecovery={() => { setEvaluationsOpen(false); setWorkersOpen(true); }} /></Modal>}
{toast && <div className="toast" role="status"><span>{toast}</span><IconButton label="关闭提示" onClick={() => setToast('')}><X size={14} /></IconButton></div>}
</div>
);
+69
View File
@@ -0,0 +1,69 @@
.evaluation-console { --ev-ink:#25344a; --ev-muted:#596a7d; --ev-line:#ccd5df; --ev-wash:#f4f7fa; color:var(--ev-ink);display:grid;gap:22px;min-width:0;overflow-wrap:anywhere; }
.evaluation-console h2,.evaluation-console h3,.evaluation-console h4,.evaluation-report h3,.evaluation-report h4 { margin:0; }
.evaluation-heading,.evaluation-report-heading { display:flex;align-items:flex-start;justify-content:space-between;gap:18px; }
.evaluation-heading h2 {font-size:21px;letter-spacing:-.025em;line-height:1.4;}
.evaluation-heading p:not(.evaluation-kicker) {font-size:13px;color:var(--ev-muted);line-height:1.7;margin:7px 0 0;}
.evaluation-kicker {font-size:12px;font-weight:650;color:#385e89;margin:0 0 6px;}
.evaluation-heading >button {flex-shrink:0;}
.evaluation-history {display:grid;grid-template-columns:auto minmax(0,1fr);align-items:center;gap:14px;font-size:13px;}
.evaluation-console input,.evaluation-console textarea,.evaluation-console select {width:100%;min-width:0;border:1px solid #bbc7d5;border-radius:5px;background:#fff;padding:9px 11px;font:inherit;font-size:13px;color:var(--ev-ink);}
.evaluation-console textarea {resize:vertical;line-height:1.65;}
.evaluation-console :focus-visible {outline:2px solid #346daa;outline-offset:3px;}
.evaluation-console button,.evaluation-report button {min-height:36px;}
.evaluation-form {display:grid;gap:18px;}
.evaluation-form label {display:grid;gap:7px;font-size:13px;font-weight:600;}
.evaluation-form fieldset {min-width:0;display:grid;gap:14px;border:1px solid var(--ev-line);border-left:3px solid #456d99;padding:18px 20px;border-radius:5px;}
.evaluation-form legend {font-size:12px;font-weight:650;padding:0 7px;}
.evaluation-budget {display:grid;grid-template-columns:repeat(4,minmax(0,1fr));gap:14px;}
.evaluation-note {font-size:12px;color:var(--ev-muted,#596a7d);line-height:1.75;margin:0;}
.evaluation-error {background:#fff5e9;border:1px solid #d9b992;border-left:3px solid #986024;color:#76501f;padding:12px 15px;font-size:13px;line-height:1.7;}
.evaluation-actions {display:flex;flex-wrap:wrap;align-items:center;gap:10px;}
.evaluation-source {border-bottom:1px solid var(--ev-line);padding:0 0 20px;}
.evaluation-source>div {display:flex;justify-content:space-between;gap:12px;align-items:center;}
.evaluation-source>p {font-size:13px;line-height:1.7;}
.evaluation-status {font-size:12px;background:#e9eff6;color:#345579;padding:5px 9px;border-radius:4px;flex-shrink:0;}
.evaluation-console dl,.evaluation-report dl {display:grid;gap:9px;margin:14px 0;font-size:12px;}
.evaluation-console dl>div,.evaluation-report dl>div {display:grid;grid-template-columns:100px minmax(0,1fr);gap:8px;}
.evaluation-console dt,.evaluation-report dt {color:#596a7d;}
.evaluation-console dd,.evaluation-report dd {margin:0;display:flex;align-items:center;gap:6px;flex-wrap:wrap;}
.evaluation-workspace {display:grid;grid-template-columns:190px minmax(0,1fr);gap:24px;align-items:start;}
.evaluation-workspace>nav {display:grid;border:1px solid var(--ev-line);border-radius:5px;overflow:hidden;}
.evaluation-workspace>nav>button {text-align:left;display:grid;gap:7px;background:#fff;border:0;border-bottom:1px solid var(--ev-line);padding:15px 13px;font:inherit;color:var(--ev-ink);cursor:pointer;}
.evaluation-workspace>nav>button:last-child {border-bottom:0;}
.evaluation-workspace>nav>button[aria-pressed=true] {background:#eaf0f7;box-shadow:inset 3px 0 #396b9e;}
.evaluation-workspace>nav small {font-size:11px;color:var(--ev-muted);}
.evaluation-workspace>nav strong {font-size:13px;line-height:1.5;}
.evaluation-workspace>nav span {font-size:11px;color:var(--ev-muted);line-height:1.6;}
.evaluation-sample {min-width:0;display:grid;gap:17px;}
.evaluation-sample>h3 {font-size:18px;line-height:1.5;}
.evaluation-goal {font-size:14px;line-height:1.8;white-space:pre-wrap;margin:0;}
.evaluation-criteria {border-left:2px solid #7f94ae;padding-left:12px;font-size:13px;line-height:1.7;}
.evaluation-criteria p {margin:5px 0 0;white-space:pre-wrap;}
.evaluation-paths {display:flex;gap:7px;flex-wrap:wrap;font-size:12px;}
.evaluation-paths code {background:var(--ev-wash);padding:4px 6px;}
.evaluation-commands {border-top:1px solid var(--ev-line);border-bottom:1px solid var(--ev-line);padding:13px 0;font-size:12px;}
.evaluation-commands summary {cursor:pointer;font-weight:650;line-height:1.7;}
.evaluation-commands>div {margin-top:13px;}
.evaluation-console pre,.evaluation-report pre {white-space:pre-wrap;overflow-wrap:anywhere;font-size:12px;line-height:1.7;background:#f1f4f8;color:#304056;padding:12px;border-radius:4px;max-height:360px;overflow:auto;}
.evaluation-lanes {display:grid;grid-template-columns:repeat(2,minmax(0,1fr));gap:0;border:1px solid #ccd5df;border-radius:5px;overflow:hidden;}
.evaluation-lanes>section {min-width:0;padding:16px;background:#fff;}
.evaluation-lanes>section+section {border-left:1px solid #ccd5df;background:#f5f8fc;}
.evaluation-lanes>section>p {font-size:12px;line-height:1.75;color:#596a7d;}
.evaluation-lane-title {display:flex;flex-wrap:wrap;gap:8px;align-items:center;justify-content:space-between;}
.evaluation-lane-title h4 {font-size:14px;}
.evaluation-lane-title>span {font-size:11px;color:#596a7d;}
.evaluation-delivery {padding:18px;border:1px solid var(--ev-line);border-left:3px solid #3d6e96;background:#f5f8fc;font-size:13px;line-height:1.8;}
.evaluation-delivery h3 {font-size:14px;}
.evaluation-confirm {display:grid;gap:18px;font-size:14px;line-height:1.8;padding:22px;}
.evaluation-report {display:grid;gap:22px;min-width:0;padding:24px;color:#25344a;overflow-wrap:anywhere;}
.evaluation-report-heading h3 {font-size:18px;}
.evaluation-report-heading p {font-size:13px;line-height:1.7;}
.evaluation-outcome {font-size:12px;color:#73551c;}
.evaluation-outcome.passed {color:#216245;}
.evaluation-outcome.failed {color:#a33e39;}
.evaluation-attempt {border-top:1px solid #ccd5df;padding:13px 0 0;margin-top:13px;font-size:12px;line-height:1.7;}
.evaluation-attempt>div {display:grid;gap:4px;margin-top:10px;}
.evaluation-attempt code {font-size:11px;color:#596a7d;}
.evaluation-report summary {font-size:12px;cursor:pointer;}
@media(max-width:760px) { .evaluation-workspace {grid-template-columns:minmax(0,1fr);gap:18px;}.evaluation-workspace>nav {grid-template-columns:repeat(3,minmax(0,1fr));}.evaluation-workspace>nav>button {padding:12px 10px;border-bottom:0;border-right:1px solid var(--ev-line);}.evaluation-budget{grid-template-columns:repeat(2,minmax(0,1fr));} }
@media(max-width:520px) { .evaluation-heading h2{font-size:18px;}.evaluation-heading{gap:10px;}.evaluation-history{grid-template-columns:minmax(0,1fr);gap:8px;}.evaluation-workspace>nav{grid-template-columns:minmax(0,1fr);}.evaluation-workspace>nav>button{border-right:0;border-bottom:1px solid var(--ev-line);}.evaluation-source>div{align-items:flex-start;flex-direction:column;}.evaluation-lanes{grid-template-columns:minmax(0,1fr);}.evaluation-lanes>section+section{border-left:0;border-top:1px solid #ccd5df;}.evaluation-report{padding:16px;}.evaluation-report-heading{flex-direction:column;}.evaluation-console button,.evaluation-report button{min-height:42px;}.evaluation-form fieldset{padding:15px;}.evaluation-console dl>div,.evaluation-report dl>div{grid-template-columns:88px minmax(0,1fr);} }
+98
View File
@@ -0,0 +1,98 @@
import { useCallback, useEffect, useRef, useState } from 'react';
import { ArrowUpRight, ClipboardCheck, Download, GitBranch, LoaderCircle, Plus, RefreshCw, Square } from 'lucide-react';
import { api, downloadBlob } from './api';
import { Modal } from './components';
import type { Config, Session, EvaluationBatch, EvaluationSample, EvaluationReport, EvaluationRole } from './types';
import './EvaluationPanel.css';
const live = ['preparing','baseline_running','candidate_running','collecting','adopting'];
const labels: Record<string,string> = {pending:'未执行',planned:'计划已封存',ready:'同源工作区就绪',preparing:'准备同源工程',baseline_running:'原版检查中',baseline_complete:'原版结果已保存',candidate_running:'候选执行中',awaiting_review:'等待改动审查',awaiting_approval:'等待检查审批',candidate_complete:'候选完成',paused:'候选已暂停',failed:'失败',stopped:'已停止',unknown:'结果待核对',sealed:'报告已封存',between_samples:'等待下一样本',completed:'本批已封存',retained:'已保留现场结束',collecting:'核对原证据',adopting:'准备采纳',awaiting_delivery:'等待父工程交付',passed:'通过'};
const label = (status:string) => labels[status] ?? status;
const split = (value:string) => value.split(/[\n,,]/).map(value=>value.trim()).filter(Boolean);
const draft = () => ({label:'',goal:'',criteria:'',paths:'',checks:'test, build'});
const number = (value:number) => value.toLocaleString('zh-CN');
const elapsed = (role:EvaluationRole) => role.startedAt && role.finishedAt ? Math.max(0,Math.round((Date.parse(role.finishedAt)-Date.parse(role.startedAt))/1000))+' 秒' : '未记录完整起止';
export function EvaluationPanel({session,config,busy,onSession,onNavigate,onReview,onRecovery}: {
session:Session;config:Config;busy:boolean;onSession:(session:Session)=>void;
onNavigate:(projectId:string,sessionId:string,view?:'task'|'changes'|'approval')=>Promise<void>;onReview:()=>void;onRecovery:()=>void;
}) {
const [batches,setBatches]=useState<EvaluationBatch[]>([]),[selected,setSelected]=useState(''),[sampleId,setSampleId]=useState('');
const [loading,setLoading]=useState(true),[action,setAction]=useState(''),[error,setError]=useState(''),[creating,setCreating]=useState(false);
const [title,setTitle]=useState(''),[samples,setSamples]=useState([draft()]);
const [evidenceMode,setEvidenceMode]=useState('configured-model');
const [tokens,setTokens]=useState(config.maxRunTokens),[steps,setSteps]=useState(Math.min(200,config.maxSteps*2)),[tools,setTools]=useState(96),[minutes,setMinutes]=useState(Math.floor(config.maxRunDurationMs/60000));
const [confirmation,setConfirmation]=useState<{batchId:string;sampleId:string;operation:string}|null>(null),[report,setReport]=useState<EvaluationReport|null>(null);
const scope=useRef(0),fetching=useRef(false),currentBatch=useRef(''),currentSample=useRef('');
const load=useCallback(async()=>{
if(fetching.current)return;fetching.current=true;const generation=scope.current;
try{const view=await api<{batches:EvaluationBatch[]}>('/sessions/'+session.id+'/evaluations');if(generation===scope.current)setBatches(view.batches);}
catch(error){if(generation===scope.current)setError(error instanceof Error?error.message:String(error));}
finally{if(generation===scope.current){fetching.current=false;setLoading(false);}}
},[session.id]);
useEffect(()=>{scope.current++;fetching.current=false;setBatches([]);setSelected('');setSampleId('');setReport(null);void load();return()=>{scope.current++;};},[load]);
useEffect(()=>{void load();},[session.revision,load]);
const batch=batches.find(batch=>batch.id===selected)??batches.at(-1),sample=batch?.samples.find(sample=>sample.id===sampleId)??batch?.samples.find(sample=>!sample.report)??batch?.samples.at(-1);
currentBatch.current=batch?.id??'';
currentSample.current=sample?.id??'';
const running=!!batch&&live.includes(batch.status);
useEffect(()=>{if(!running)return;const timer=setInterval(()=>void load(),1200);return()=>clearInterval(timer);},[running,load]);
useEffect(()=>{setConfirmation(null);setReport(null);setError('');},[session.id,batch?.id,sample?.id]);
async function execute(operation:string,target=batch,item=sample){
if(!target||currentBatch.current!==target.id)return;const generation=scope.current;setAction(operation);setError('');setConfirmation(null);
try{const reply=await api<{session:Session}>('/sessions/'+session.id+'/evaluations/'+target.id+'/'+operation,{fingerprint:target.fingerprint,...(item?{sampleId:item.id}:{})});if(generation===scope.current){onSession(reply.session);await load();}}
catch(error){if(generation===scope.current)setError(error instanceof Error?error.message:String(error));}
finally{if(generation===scope.current)setAction('');}
}
async function create(){const generation=scope.current;setAction('create');setError('');
try{const reply=await api<{batch:EvaluationBatch;session:Session}>('/sessions/'+session.id+'/evaluations',{title,samples:samples.map(sample=>({...sample,paths:split(sample.paths),checks:split(sample.checks)})),budget:{maxTokens:tokens,maxSteps:steps,maxToolCalls:tools,maxDurationMs:minutes*60000},evidenceMode});if(generation===scope.current){onSession(reply.session);setSelected(reply.batch.id);setCreating(false);await load();}}
catch(error){if(generation===scope.current)setError(error instanceof Error?error.message:String(error));}
finally{if(generation===scope.current)setAction('');}
}
async function showReport(){if(!batch||!sample)return;const generation=scope.current,target=batch.id,id=sample.id;setAction('report');setError('');
try{const result=await api<EvaluationReport>('/sessions/'+session.id+'/evaluations/'+batch.id+'/report?sampleId='+sample.id);if(generation===scope.current&&currentBatch.current===target&&currentSample.current===id)setReport(result);}
catch(error){if(generation===scope.current)setError(error instanceof Error?error.message:String(error));}finally{if(generation===scope.current)setAction('');}
}
const confirm=(operation:string)=>{if(batch&&sample)setConfirmation({batchId:batch.id,sampleId:sample.id,operation});};
const canCreate=!busy&&!action&&(!batch||['completed','retained'].includes(batch.status));
return <div className="evaluation-console" aria-label="工程评测控制台">
<header className="evaluation-heading"><div><p className="evaluation-kicker">工程质量 · 固定样本</p><h2>同一份需求,核对改进前后</h2><p>封存来源与原测试,在隔离工程里生成候选,再把真实结果带回审查。</p></div><button className="button secondary" disabled={loading||!!action} onClick={()=>void load()}><RefreshCw size={14}/>刷新</button></header>
{error&&<div className="evaluation-error" role="alert">{error}</div>}
{loading?<p role="status"><LoaderCircle size={15} className="spin"/>正在读取评测…</p>:<>
{!!batches.length&&<label className="evaluation-history">评测记录<select aria-label="选择评测记录" value={batch?.id??''} onChange={event=>{setSelected(event.target.value);setSampleId('');setCreating(false);}}>{batches.map(batch=><option key={batch.id} value={batch.id}>{batch.title} · {label(batch.status)}</option>)}</select></label>}
{(!batch||creating)?<section className="evaluation-form" aria-label="新建工程评测">
<label>评测名称<input value={title} maxLength={120} onChange={event=>setTitle(event.target.value)} placeholder="例如 修复导入边界并保持原测试"/></label>
<label>评测来源<select aria-label="评测来源" value={evidenceMode} onChange={event=>setEvidenceMode(event.target.value)}><option value="configured-model">实际配置模型 · 保留真实结果</option><option value="local-fixture">本地固定响应 · 只验证执行链</option></select></label>
<p className="evaluation-note">每批 1–3 个样本,按队列显式推进。每个样本都从当前已提交 HEAD 检出;原暂存和未提交内容留在来源工程。</p>
{samples.map((sample,index)=><fieldset key={index}><legend>固定样本 {index+1}</legend>{(['label','goal','criteria','paths','checks'] as const).map(key=><label key={key}>{({label:'样本名称',goal:'具体需求',criteria:'验收条件',paths:'允许修改的源码文件',checks:'完整检查名称'})[key]}{key==='label'?<input value={sample[key]} maxLength={64} onChange={event=>setSamples(values=>values.map((value,i)=>i===index?{...value,[key]:event.target.value}:value))}/>:<textarea aria-label={({label:'样本名称',goal:'具体需求',criteria:'验收条件',paths:'允许修改的源码文件',checks:'完整检查名称'})[key]} rows={key==='goal'||key==='criteria'?3:2} maxLength={key==='goal'?6000:4000} value={sample[key]} placeholder={key==='paths'?'例如 src/parser.mjs,每行一个已有文件':key==='checks'?'test, typecheck, build;Python 使用 python:test, python:compile':undefined} onChange={event=>setSamples(values=>values.map((value,i)=>i===index?{...value,[key]:event.target.value}:value))}/>}</label>)}{samples.length>1&&<button className="text-button" onClick={()=>setSamples(values=>values.filter((_,i)=>i!==index))}>移除此未保存样本</button>}</fieldset>)}
{samples.length<3&&<button className="button secondary" onClick={()=>setSamples(values=>[...values,draft()])}><Plus size={14}/>添加固定样本</button>}
<div className="evaluation-budget"><label>总 token 预算<input type="number" min={samples.length*2000} max={Math.min(config.maxTaskTokens,config.maxRunTokens*samples.length)} value={tokens} onChange={event=>setTokens(Number(event.target.value))}/></label><label>总请求轮次<input type="number" min={samples.length*2} max={200} value={steps} onChange={event=>setSteps(Number(event.target.value))}/></label><label>总工具次数<input type="number" min={samples.length*2} max={512} value={tools} onChange={event=>setTools(Number(event.target.value))}/></label><label>总时限(分钟)<input type="number" min={1} max={Math.floor(config.maxRunDurationMs/60000)} value={minutes} onChange={event=>setMinutes(Number(event.target.value))}/></label></div>
<p className="evaluation-note">预算按样本均分,续接保留累计用量。总时限从首次原版检查开始,包含审查等待。候选不修改测试、脚本或运行配置。</p>
<div className="evaluation-actions"><button className="button" disabled={!canCreate||!title.trim()||samples.some(sample=>!sample.label.trim()||!sample.goal.trim()||!sample.criteria.trim()||!split(sample.paths).length||!split(sample.checks).length)} onClick={()=>void create()}><ClipboardCheck size={15}/>封存评测计划</button>{creating&&<button className="button secondary" onClick={()=>setCreating(false)}>返回原评测</button>}</div>
</section>:sample&&<>
<section className="evaluation-source"><div><strong>{batch.title}</strong><span className="evaluation-status">{label(batch.status)}</span></div><p>{batch.message}</p><dl><div><dt>固定来源</dt><dd><GitBranch size={13}/>{batch.source.branch} · <code>{batch.source.head.slice(0,12)}</code></dd></div><div><dt>候选模型</dt><dd>{batch.model.mode==='demo'?'离线演示,不能验证模型能力':batch.model.model}</dd></div><div><dt>完整样本分母</dt><dd>{batch.summary.finished} / {batch.summary.samples} 已封存 · 通过 {batch.summary.candidate.passed} · 失败 {batch.summary.candidate.failed} · 未知 {batch.summary.candidate.unknown} · 未执行 {batch.summary.candidate.pending}</dd></div></dl><p className="evaluation-note">{batch.evidenceMode==='local-fixture'?'本批使用本地固定响应,仅验证执行链和真实检查。':batch.summary.scope}</p></section>
<div className="evaluation-workspace"><nav aria-label="固定样本队列">{batch.samples.map((item,index)=><button key={item.id} aria-pressed={sample.id===item.id} onClick={()=>{setSampleId(item.id);setReport(null);}}><small>样本 {index+1}</small><strong>{item.label}</strong><span>{label(item.status)}</span>{item.report&&<span>原版 {label(item.report.summary.original)} → 候选 {label(item.report.summary.candidate)}</span>}</button>)}</nav>
<article className="evaluation-sample"><h3>{sample.label}</h3><p className="evaluation-goal">{sample.goal}</p><div className="evaluation-criteria"><strong>验收条件</strong><p>{sample.criteria}</p></div><div className="evaluation-paths">{sample.paths.map(path=><code key={path}>{path}</code>)}</div>
{sample.error&&<p className="evaluation-error">{sample.error}</p>}
<details className="evaluation-commands" open={sample.status==='ready'}><summary>固定检查 · {sample.checks.join(' / ')}</summary>{sample.commands?.map(command=><div key={command.name}><strong>{command.command}</strong><pre>{command.script}</pre>{command.lifecycle?.map(step=><p key={step.name}>{step.name}:<code>{step.script}</code></p>)}</div>)??<p>工作区准备完成后显示实际命令,执行前需要核对。</p>}</details>
{!!sample.workers?.length&&<div className="evaluation-lanes">{sample.workers.map(worker=><section key={worker.id} aria-label={worker.evaluation?.role==='original'?'原版工程':'候选工程'}><div className="evaluation-lane-title"><h4>{worker.evaluation?.role==='original'?'原版':'候选'}</h4><span>{worker.evaluation?.role==='original'?'只跑固定检查':label(worker.sessionStatus==='awaiting_approval'?'awaiting_approval':worker.status)}</span></div><p>{worker.summary}</p>{worker.status!=='unprepared'&&<button className="button secondary" onClick={()=>void onNavigate(worker.projectId,worker.sessionId,worker.sessionStatus==='awaiting_approval'?'approval':worker.sessionStatus==='needs_review'?'changes':'task')}>{worker.sessionStatus==='awaiting_approval'?'核对并审批检查':worker.sessionStatus==='needs_review'?'审查候选改动':'打开原任务'}<ArrowUpRight size={13}/></button>}</section>)}</div>}
{sample.contractSha256&&<p className="evaluation-note">同源输入已封存 · SHA-256 <code>{sample.contractSha256.slice(0,16)}</code></p>}
<div className="evaluation-actions">
{sample.status==='pending'&&<button className="button" disabled={!!action||running||busy||batch.samples.slice(0,batch.samples.indexOf(sample)).some(sample=>!sample.report)} onClick={()=>confirm('prepare')}>准备同源工作区</button>}
{sample.status==='ready'&&<button className="button" disabled={!!action||running} onClick={()=>confirm('baseline')}>核对并运行原版</button>}
{['baseline_complete','paused','awaiting_review','failed','stopped'].includes(sample.status)&&!!sample.baselineJobId&&!sample.report&&<button className="button" disabled={!!action||running||config.mode==='demo'||sample.workers?.some(worker=>worker.evaluation?.role==='candidate'&&worker.sessionStatus==='needs_review')} onClick={()=>confirm('start')}>{sample.status==='baseline_complete'?'开始候选改进':'继续候选任务'}</button>}
{running&&<button className="button secondary" disabled={action==='stop'} onClick={()=>void execute('stop')}><Square size={14}/>停止本批执行</button>}
{!running&&!sample.report&&sample.status!=='pending'&&<button className="button secondary" disabled={!!action} onClick={()=>confirm('seal')}>封存当前结果</button>}
{sample.report&&<button className="button secondary" disabled={!!action} onClick={()=>void showReport()}>查看原版 / 候选对照</button>}
{sample.report?.summary.candidate==='passed'&&!batch.adoption&&batch.samples.every(sample=>sample.report)&&<button className="button" disabled={!!action||running||batch.status==='retained'} onClick={()=>confirm('adopt')}>采纳此候选,准备父工程审查</button>}
{batch.status==='unknown'&&<button className="button secondary" onClick={onRecovery}>打开原协作核对现场</button>}
</div>
</article></div>
{batch.adoption&&<section className="evaluation-delivery"><h3>父工程交付</h3><p>{batch.adoption.delivery?'已完成本地提交 '+batch.adoption.delivery.commit:batch.message}</p>{batch.status==='awaiting_delivery'&&<div className="evaluation-actions"><button className="button secondary" onClick={onReview}>审查父工程变更集</button><button className="button secondary" disabled={!!action||busy} onClick={()=>void execute('finalize')}>核对父工程检查与 Git</button></div>}</section>}
<footer className="evaluation-actions">{!running&&!['completed','retained'].includes(batch.status)&&<button className="text-button" disabled={!!action} onClick={()=>confirm('retain')}>保留现场,结束本批</button>}{canCreate&&<button className="button secondary" onClick={()=>setCreating(true)}><Plus size={14}/>新建下一批评测</button>}</footer>
</>}
</>}
{confirmation&&batch&&sample&&<Modal title="确认评测操作" onClose={()=>setConfirmation(null)}><div className="evaluation-confirm"><h3>{sample.label}</h3><p>{({prepare:'从封存 HEAD 创建两个独立 Git 工作区,保留来源原修改。',baseline:'请核对上方完整检查命令。将以本机账户执行原工程程序,不请求模型。',start:'将请求封存模型并使用剩余预算。候选提案仍需审查,工程检查仍需审批。',seal:'保存当前全部尝试、失败、停止与未知。封存后该样本不能再执行或修改。',adopt:'核对原通过报告及当前来源后,生成父工程待审查变更集。应用、父检查和 Git 交付仍需确认。',retain:'保留原任务、工作区和全部结果,结束本批;未执行样本仍计入原分母。'} as Record<string,string>)[confirmation.operation]}</p><button className="button" disabled={!!action} onClick={()=>{if(confirmation.batchId===batch.id&&confirmation.sampleId===sample.id)void execute(confirmation.operation);}}>确认{({prepare:'准备',baseline:'运行原版',start:'开始候选',seal:'封存结果',adopt:'准备采纳',retain:'保留现场结束'} as Record<string,string>)[confirmation.operation]}</button></div></Modal>}
{report&&sample&&<Modal title="原版与候选对照" wide onClose={()=>setReport(null)}><div className="evaluation-report"><div className="evaluation-report-heading"><div><h3>{report.label}</h3>{report.evidenceMode==='local-fixture'&&<p className="evaluation-note">本地固定响应 · 执行链验收</p>}<p>{report.inputVerified?'原测试与非候选输入已核对':report.inputError??'原输入无法核对'}</p></div><button className="button secondary" onClick={()=>downloadBlob(new Blob([JSON.stringify(report,null,2)],{type:'application/json'}),'agent-workbench-evaluation-'+report.sampleId+'.json')}><Download size={14}/>导出原报告</button></div><div className="evaluation-lanes">{(['original','candidate'] as const).map(role=>{const result=report[role];return <section key={role}><div className="evaluation-lane-title"><h4>{role==='original'?'原版':'候选'}</h4><strong className={'evaluation-outcome '+result.outcome}>{label(result.outcome)}</strong></div><p>{result.reason??result.evidenceError}</p><dl><div><dt>用时</dt><dd>{elapsed(result)}</dd></div><div><dt>模型请求</dt><dd>{result.usage?.requests??'未知'}</dd></div><div><dt>上游用量</dt><dd>{result.usage?.reported?number(result.usage.totalTokens)+' tokens':'未知;保留预算估算'}</dd></div><div><dt>费用</dt><dd>未报告</dd></div><div><dt>实际人工操作</dt><dd>审批 {result.approvals?.filter(approval=>approval.status==='approved').length??0} 次 · 应用提案 {result.changes?.filter(change=>change.status==='applied').length??0} 项</dd></div></dl>{result.attempts.length?result.attempts.map((attempt,index)=><div className="evaluation-attempt" key={attempt.id}><strong>检查尝试 {index+1} · {label(attempt.status)}</strong>{attempt.results.map(check=><div key={check.logArtifactId}><span>{check.name} · {check.status?label(check.status):'证据缺失'}</span>{check.tests&&<span>测试 {check.tests.passed}/{check.tests.total} · 失败 {check.tests.failed}</span>}<code>{check.logSha256?.slice(0,16)??'原日志无法核对'}</code></div>)}</div>):<p>没有完整检查结果;保持未知。</p>}</section>;})}</div><p className="evaluation-note">{report.note}</p><details><summary>报告原字段与工件引用</summary><pre>{JSON.stringify(report,null,2)}</pre></details></div></Modal>}
</div>;
}
+7 -7
View File
@@ -104,16 +104,16 @@ export function WorkerPanel({ session, config, busy, onSession, onNavigate, onRe
</div>}
{group && !creating && <>
{group.children[0]?.engine?.strategy && <p className="worker-engine-badge">本计划的 Purge 提示实验 · {group.children[0].engine.strategy.source.version} · {group.children[0].engine.strategy.sha256.slice(0, 16)}</p>}
<section className="worker-parent"><div className="worker-parent-top"><span className={'worker-status ' + (group.requiresVerification || group.status === 'conflict' ? 'attention' : '')}>{statusLabel(group.status)}</span><code>{group.source.identity.branch} · {group.source.identity.head.slice(0, 12)}</code></div><h3>{group.goal}</h3><p>{group.summary}</p><dl><div><dt>模型</dt><dd>{group.model.mode === 'demo' ? '离线演示' : group.model.model}</dd></div><div><dt>执行方式</dt><dd>{group.children[0].engine ? "DeepSeek Harness " + group.children[0].engine.version : "工作台受限子任务"}</dd></div><div><dt>总预算</dt><dd>{number(group.budget.maxTokens)} tokens / {group.budget.maxSteps} 轮 / {group.budget.maxToolCalls} 次工具</dd></div><div><dt>累计用量</dt><dd>{number(group.usage?.accountedTokens)} tokens / {group.usage?.steps ?? 0} 轮 / {group.usage?.toolCalls ?? 0} 次工具 / {group.usage?.requests ?? 0} 次模型请求</dd></div><div><dt>原修改</dt><dd>{group.source.changedPaths.length ? group.source.changedPaths.map(value => value.path).join('、') : '无未提交修改'}</dd></div></dl></section>
<div className={"worker-lanes" + (group.children.length === 1 ? " single" : "")}>{(group.workers ?? []).map((child, index) => <section className="worker-lane" key={child.id}><div className="worker-lane-title"><span>通道 {index + 1}</span><span className={'worker-status ' + (child.requiresVerification ? 'attention' : '')}>{statusLabel(child.sessionStatus === 'awaiting_approval' ? 'awaiting_approval' : child.status)}</span></div><h3>{child.label}</h3>{child.engine && <p className="worker-engine-badge">DeepSeek Harness {child.engine.version} · 固定安装 {child.engine.controlledSha256?.slice(0, 12)}</p>}<p className="worker-goal">{child.goal}</p><div className="worker-scope"><span>允许修改</span>{child.paths.map(path => <code key={path}>{path}</code>)}</div><p className="worker-meter">{number(child.usage?.accountedTokens)} / {number(child.budget.maxTokens)} tokens · {child.usage?.steps ?? 0}/{child.budget.maxSteps} 轮</p><p className="worker-summary">{child.summary ?? '先创建独立工作区,再开始子任务。'}</p>{child.result && <div className="worker-proof"><Check size={14} />{child.result.paths.length} 个原改动 · 当前检查已封存</div>}<div className="worker-actions">{child.status !== 'unprepared' && <button className="button secondary" onClick={() => void onNavigate(child.projectId, child.sessionId, child.sessionStatus === "awaiting_approval" ? "approval" : child.sessionStatus === "needs_review" ? "changes" : "task")}>{child.sessionStatus === 'awaiting_approval' ? '查看并审批检查' : child.sessionStatus === 'needs_review' ? '审查子任务改动' : '打开子任务'}<ArrowUpRight size={13} /></button>}{['ready', 'paused', 'failed', 'awaiting_review'].includes(child.status) && !running && child.sessionStatus !== 'needs_review' && <button className="text-button" disabled={!!action || config.mode === 'demo' || group.requiresVerification} onClick={() => setConfirmation({ group, operation: 'start', workerId: child.id })}><Play size={13} />继续子任务</button>}</div></section>)}</div>
{group.evaluation && <p className="worker-note">本协作属于固定工程评测。执行与采纳请使用「工程评测」;这里保留原任务导航和中断现场核对。</p>}<section className="worker-parent"><div className="worker-parent-top"><span className={'worker-status ' + (group.requiresVerification || group.status === 'conflict' ? 'attention' : '')}>{statusLabel(group.status)}</span><code>{group.source.identity.branch} · {group.source.identity.head.slice(0, 12)}</code></div><h3>{group.goal}</h3><p>{group.summary}</p><dl><div><dt>模型</dt><dd>{group.model.mode === 'demo' ? '离线演示' : group.model.model}</dd></div><div><dt>执行方式</dt><dd>{group.children[0].engine ? "DeepSeek Harness " + group.children[0].engine.version : "工作台受限子任务"}</dd></div><div><dt>总预算</dt><dd>{number(group.budget.maxTokens)} tokens / {group.budget.maxSteps} 轮 / {group.budget.maxToolCalls} 次工具</dd></div><div><dt>累计用量</dt><dd>{number(group.usage?.accountedTokens)} tokens / {group.usage?.steps ?? 0} 轮 / {group.usage?.toolCalls ?? 0} 次工具 / {group.usage?.requests ?? 0} 次模型请求</dd></div><div><dt>原修改</dt><dd>{group.source.changedPaths.length ? group.source.changedPaths.map(value => value.path).join('、') : '无未提交修改'}</dd></div></dl></section>
<div className={"worker-lanes" + (group.children.length === 1 ? " single" : "")}>{(group.workers ?? []).map((child, index) => <section className="worker-lane" key={child.id}><div className="worker-lane-title"><span>通道 {index + 1}</span><span className={'worker-status ' + (child.requiresVerification ? 'attention' : '')}>{statusLabel(child.sessionStatus === 'awaiting_approval' ? 'awaiting_approval' : child.status)}</span></div><h3>{child.label}</h3>{child.engine && <p className="worker-engine-badge">DeepSeek Harness {child.engine.version} · 固定安装 {child.engine.controlledSha256?.slice(0, 12)}</p>}<p className="worker-goal">{child.goal}</p><div className="worker-scope"><span>允许修改</span>{child.paths.map(path => <code key={path}>{path}</code>)}</div><p className="worker-meter">{number(child.usage?.accountedTokens)} / {number(child.budget.maxTokens)} tokens · {child.usage?.steps ?? 0}/{child.budget.maxSteps} 轮</p><p className="worker-summary">{child.summary ?? '先创建独立工作区,再开始子任务。'}</p>{child.result && <div className="worker-proof"><Check size={14} />{child.result.paths.length} 个原改动 · 当前检查已封存</div>}<div className="worker-actions">{child.status !== 'unprepared' && <button className="button secondary" onClick={() => void onNavigate(child.projectId, child.sessionId, child.sessionStatus === "awaiting_approval" ? "approval" : child.sessionStatus === "needs_review" ? "changes" : "task")}>{child.sessionStatus === 'awaiting_approval' ? '查看并审批检查' : child.sessionStatus === 'needs_review' ? '审查子任务改动' : '打开子任务'}<ArrowUpRight size={13} /></button>}{!group.evaluation && ['ready', 'paused', 'failed', 'awaiting_review'].includes(child.status) && !running && child.sessionStatus !== 'needs_review' && <button className="text-button" disabled={!!action || config.mode === 'demo' || group.requiresVerification} onClick={() => setConfirmation({ group, operation: 'start', workerId: child.id })}><Play size={13} />继续子任务</button>}</div></section>)}</div>
<section className="worker-review"><div className="worker-review-title"><GitBranch size={18} /><h3>父工程成果审查</h3></div>{group.collection ? <><p>{group.collection.paths.length} 个路径已核对,原记录 <code>{group.collection.sha256.slice(0, 16)}</code></p>{group.collection.conflicts.length > 0 && <ul className="worker-conflicts">{group.collection.conflicts.map((conflict, i) => <li key={i}><code>{conflict.path}</code><span>{conflict.reason}</span></li>)}</ul>}<button className="text-button" disabled={!!action} onClick={() => void showRecord(group, 'collection')}>查看原成果记录</button></> : <p>等待 {group.children.length} 个子任务完成实际修改与检查。重叠路径、来源变化或原证据失效会阻止合并。</p>}{group.merge?.changeSetId && <button className="button secondary" onClick={onReview}>打开父工程变更集审查<ArrowUpRight size={14} /></button>}</section>
<div className="worker-actions worker-primary-actions">
{['planned', 'partially_prepared'].includes(group.status) && <button className="button" disabled={!!action || busy} onClick={() => setConfirmation({ group, operation: 'prepare' })}><GitBranch size={15} />{group.children.length === 2 ? '创建两个工作区' : '创建引擎工作区'}</button>}
{['ready', 'paused', 'awaiting_review'].includes(group.status) && <button className="button" disabled={!!action || running || config.mode === 'demo' || group.requiresVerification || group.workers?.some(child => child.sessionStatus === 'needs_review')} onClick={() => setConfirmation({ group, operation: 'start' })}><Play size={15} />{group.status === 'ready' ? group.children.length === 2 ? '开始两个子任务' : '开始引擎任务' : '继续未完成子任务'}</button>}
{!group.evaluation && ['planned', 'partially_prepared'].includes(group.status) && <button className="button" disabled={!!action || busy} onClick={() => setConfirmation({ group, operation: 'prepare' })}><GitBranch size={15} />{group.children.length === 2 ? '创建两个工作区' : '创建引擎工作区'}</button>}
{!group.evaluation && ['ready', 'paused', 'awaiting_review'].includes(group.status) && <button className="button" disabled={!!action || running || config.mode === 'demo' || group.requiresVerification || group.workers?.some(child => child.sessionStatus === 'needs_review')} onClick={() => setConfirmation({ group, operation: 'start' })}><Play size={15} />{group.status === 'ready' ? group.children.length === 2 ? '开始两个子任务' : '开始引擎任务' : '继续未完成子任务'}</button>}
{running && <button className="button worker-stop" disabled={action === 'stop'} onClick={() => void execute(group, 'stop')}><Square size={14} />停止全部</button>}
{['ready_to_collect', 'collected', 'conflict'].includes(group.status) && <button className="button secondary" disabled={!!action || running} onClick={() => void execute(group, 'collect')}><ShieldCheck size={15} />核对成果</button>}
{group.status === 'collected' && <button className="button" disabled={!!action} onClick={() => void execute(group, 'merge')}><GitBranch size={15} />准备合并审查</button>}
{group.merge?.changeSetId && <button className="button secondary" disabled={!!action || busy} onClick={() => void execute(group, 'finalize')}>核对检查与 Git 交付</button>}
{!group.evaluation && ['ready_to_collect', 'collected', 'conflict'].includes(group.status) && <button className="button secondary" disabled={!!action || running} onClick={() => void execute(group, 'collect')}><ShieldCheck size={15} />核对成果</button>}
{!group.evaluation && group.status === 'collected' && <button className="button" disabled={!!action} onClick={() => void execute(group, 'merge')}><GitBranch size={15} />准备合并审查</button>}
{!group.evaluation && group.merge?.changeSetId && <button className="button secondary" disabled={!!action || busy} onClick={() => void execute(group, 'finalize')}>核对检查与 Git 交付</button>}
{group.requiresVerification && <><button className="button secondary" disabled={!!action || running} onClick={() => void execute(group, 'reconcile')}>核对原输入与现场</button>{group.evidence && <><button className="text-button" disabled={!!action} onClick={() => void showRecord(group, 'evidence')}>查看核对记录</button><button className="button secondary" disabled={!!action || readEvidenceSha !== group.evidence.sha256} onClick={() => setConfirmation({ group, operation: 'acknowledge' })}>确认核对并继续</button><button className="text-button" disabled={!!action || readEvidenceSha !== group.evidence.sha256} onClick={() => setConfirmation({ group, operation: 'acknowledge', retain: true })}>确认核对并保留现场结束</button></>}</>}
{!running && !group.requiresVerification && !['completed', 'discarded', 'resolved_external'].includes(group.status) && <button className="text-button" disabled={!!action} onClick={() => setConfirmation({ group, operation: 'discard' })}>结束协作并保留现场</button>}
{canPlan && <button className="button secondary" onClick={() => { setPurge(emptyPurgeDraft()); setCreating(true); }}>创建下一父目标</button>}
+25
View File
@@ -201,6 +201,31 @@ export interface CollaborationGroup {
evidence?: { artifactId: string; sha256: string } | null;
}
export interface WorkerView { groups: CollaborationGroup[]; worker: WorkerRecord | null }
export interface CollaborationGroup { evaluation?: { batchId: string; sampleId: string } }
export interface WorkerDefinition { evaluation?: { batchId: string; sampleId: string; role: 'original' | 'candidate' } }
export interface EvaluationRole {
outcome: string; reason?: string; evidenceError?: string; workerStatus?: string; startedAt?: string; finishedAt?: string;
usage: { requests: number; totalTokens: number; unknownRequests?: number; estimatedUnknownTokens?: number; reported: boolean; cost?: number | null } | null;
attempts: { id: string; status: string; summary?: string; results: { name: string; command?: string; status?: string; exitCode?: number; durationMs?: number; tests?: CheckResult['tests']; logArtifactId: string; logSha256?: string }[] }[];
approvals?: { id: string; status: string; decidedAt?: string }[]; changes?: { id: string; path: string; status: string }[];
}
export interface EvaluationReport {
schemaVersion: number; batchId: string; sampleId: string; label: string; goal: string; criteria: string; source: { head: string };
inputVerified: boolean; inputError?: string; summary: { original: string; candidate: string }; original: EvaluationRole; candidate: EvaluationRole;
evidenceMode: string; note: string; sealedAt: string;
}
export interface EvaluationSample {
id: string; label: string; goal: string; criteria: string; paths: string[]; checks: string[]; status: string; groupId?: string; error?: string;
baselineAttemptedAt?: string; baselineJobId?: string; contractSha256?: string;
commands?: { name: string; command: string; script: string; lifecycle?: { name: string; script: string }[] }[];
report: { artifactId: string; sha256: string; summary: { original: string; candidate: string } } | null; workers?: WorkerRecord[];
}
export interface EvaluationBatch {
id: string; title: string; fingerprint: string; status: string; source: { head: string; branch: string }; model: ModelSnapshot;
message: string; evidenceMode: string; samples: EvaluationSample[]; budget: WorkerBudget & { maxDurationMs: number }; deadline: number | null;
adoption: { sampleId: string; status: string; delivery?: { commit: string } } | null;
summary: { samples: number; finished: number; candidate: Record<string, number>; scope: string };
}
export interface McpTool { name: string; description: string; inputSchema: Record<string, unknown>; unsupported?: string }
export interface McpRegistration {
id: string; projectId: string; name: string; launch: { executable: { path: string; sha256: string; bytes: number }; entry: { path: string; sha256: string; bytes: number } | null; args: string[]; environment: string; identityScope: string };
+2 -2
View File
@@ -7,8 +7,8 @@
1. [complete] A:固定样本/基线/需求/允许文件/检查/预算契约;审查原40基准及Node/React/Python,明确夹具与真实推理边界。
2. [complete] B:复用两个隔离工作区,一个只做原版检查,一个生成候选;有界串行样本账本、原状态与停止/重启保护,不重放完成工具。
3. [complete] C:原版/候选检查、变更、审批/人工介入、用时和原用量对照;缺失与失败保留原分母,不补造评分。
4. [in_progress] D:桌面/390px评测入口、完整准备/执行/审查/比较/采纳流程,复用现有审批/差异/检查/Git。
5. [pending] E:实际Node/React/Python工程和自身模块试点,测试输入固定、来源修改保持,真实模型自主完成率单列待验。
4. [complete] D:桌面/390px评测入口、完整准备/执行/审查/比较/采纳7组通过,停止/来源变化保留原归属;实际两视口截图已回看,复用现有审批/差异/检查/Git。
5. [in_progress] E:实际Node/React/Python工程和自身模块试点,测试输入固定、来源修改保持,真实模型自主完成率单列待验。
6. [pending] F:完整回归、独立源码/便携/安装/保留状态升级、原生回看、证据/回顾/本地Git与下一计划。
## 1.19 项目上下文缓存与长任务效率(已完成)